Business Compromise Detection with Agentic AI and LLM-driven Knowledge Discovery
This study finds the autonomous agent is a strong, recall-heavy signal extractor but an unreliable final arbiter, conceding precision on ambiguous decisions, so keeps the agent as an investigator that emits a structured, interpretable signal vector, and delegates the verdict to a neuro-symbolic stage.
D. Palma, Kyu Bin Kim, Zhen Han et al.
· 0 citations